Does any mission have a propulsion system such that thrusters are aligned automatically, in order to only give delta-v, but without changing angular momentum? If so, how is this done? That is to say, how is the support of such thrusters designed, in order to ensure that the reaction force vector passes through the CoG of the spacecraft?
Any propulsion system that uses propellant is going to be hard pressed to meet this goal. As propellant is used up the centre of mass changes (the propellant contributes to the centre of mass). You could design a propulsion system with a spherical propellant tank, which is aligned with the centre of the spacecraft, thereby negating the effect of loss of fuel on the CoM, however that would still require you to design your spacecraft with zero products of inertia. If you've ever attempted a complete satellite configuration, right down to the nuts and bolts, then you will know just how hard it is to get zero's for the products of inertia (I'd suggest that it's borderline impossible to be completely zero - material inconsistencies etc.).
Two of methods do spring to mind however. First you have things like solar sails, these need no propellant from the spacecraft, and in theory as long as they are aligned perpendicular to the force vector from the sun they should experience a constant force over the surface (solar particle flux in directions of the sail length and width should be symmetrical about the centre of the sail). All this then requires is that you design your spacecraft to have a centre of mass directly aligned with the sun force vector.
The other thought that sprung to mind was a spin stabilised spacecraft. If you are performing a long duration, continuous burn you mitigate any part of the burn that is not along the preburn angular momentum vector. So if you want to go directly forward and not change angular momentum, spin around. Change your angular momentum to negate any accidental changes in your angular momentum.
I can describe the once-existing method that was used on the Shuttle Orbiter. Large orbital changes were made using the 2 Orbital Maneuvering System (OMS) engines (6000 lbf each). These engines were gimbaled and positioned by a set of electric gimbal actuators to keep them pointing through the Orbiter's CG. The gimbaling was handled by the Digital Autopilot (DAP) as described here:
For the OMS thrusting period, the orbital state (position and vector) is produced by navigation incorporating inertial measurement unit delta velocities during powered and coasting flight. This state is sent to guidance, which uses target inputs through the CRT to compute thrust direction commands and commanded attitude for flight control and thrusting parameters for CRT display. Flight control converts the commands into OMS engine gimbal angles (thrust vector control) for an automatic thrusting period. OMS thrust vector control for normal two-engine thrusting is entered by depressing the orbital DAP auto push button light indicator with both RHCs within software detents. OMS manual thrust vector control for both OMS engines is entered by depressing the orbital man DAP push button light indicator or by moving the commander's or pilot's RHC out of detent; the flight crew supplies the rate commands to the TVC system instead of guidance. The manual RHC rotation requests are proportional to RHC deflections and are converted into gimbal angles. OMS thrust in either case is applied through the spacecraft's center of gravity.
Acronym deciphering: CRT = cathode ray tube, the crew display (this manual is old), RHC = rotational hand controller aka the joystick, TVC = thrust vector control
If you want to read more about the OMS itself, that's here.